Rustup to "rustc 1.58.0-nightly (efd048394 2021-10-20)"

[nstilt1]

Fixes #19
based on #10

I was able to find 2 specific nightly commits that caused some of the changes:

• eec856b
  • this one removed sym::send_trait and switched usage to use sym::Send
• db9fea5
  • this one eliminated .krate() from hir().krate().visit_all_item_likes() and hir().krate().items()

This update might make it a little easier for the next update... but nightly rust really does change nightly.

[nstilt1]

Well. It doesn't seem to completely work. ./test.py works, but cargo rudra and docker-cargo-rudra ./ ./rudra-reports doesn't show any log files. Will try the debug version

[nstilt1]

Hmm... maybe I was just expecting a log file to be generated, but I suppose it only does that when it finds something. I tried it out on rust-sha v1.0.3 and got these results

$ cargo rudra

rust-sha$ cargo rudra
2023-12-23 16:39:49.854682 |INFO | [rudra-progress] Running cargo rudra
2023-12-23 16:39:51.728205 |INFO | [rudra-progress] Running rudra for target lib:sha
warning: use of deprecated macro `try`: use the `?` operator instead
   --> src/utils.rs:102:23
    |
102 |         let mut len = try!(self.inner.read(&mut self.buf[..]));
    |                       ^^^
    |
    = note: `#[warn(deprecated)]` on by default

warning: use of deprecated macro `try`: use the `?` operator instead
   --> src/utils.rs:108:20
    |
108 |             len += try!(reader.read_pad(&mut self.buf));
    |                    ^^^

warning: use of deprecated macro `try`: use the `?` operator instead
   --> src/utils.rs:172:9
    |
172 |         try!(Write::write(buf, &self.suffix[..]));
    |         ^^^

warning: the item `Default` is imported redundantly
   --> src/sha1.rs:330:13
    |
330 |         use std::default::Default;
    |             ^^^^^^^^^^^^^^^^^^^^^
    |
   ::: /home/somedooby/.rustup/toolchains/nightly-2021-10-21-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/std/src/prelude/mod.rs:109:13
    |
109 |     pub use super::v1::*;
    |             ------------ the item `Default` is already defined here
    |
    = note: `#[warn(unused_imports)]` on by default

warning: the item `Default` is imported redundantly
   --> src/sha256.rs:379:13
    |
379 |         use std::default::Default;
    |             ^^^^^^^^^^^^^^^^^^^^^
    |
   ::: /home/somedooby/.rustup/toolchains/nightly-2021-10-21-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/std/src/prelude/mod.rs:109:13
    |
109 |     pub use super::v1::*;
    |             ------------ the item `Default` is already defined here

warning: the item `Default` is imported redundantly
   --> src/sha512.rs:312:13
    |
312 |         use std::default::Default;
    |             ^^^^^^^^^^^^^^^^^^^^^
    |
   ::: /home/somedooby/.rustup/toolchains/nightly-2021-10-21-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/std/src/prelude/mod.rs:109:13
    |
109 |     pub use super::v1::*;
    |             ------------ the item `Default` is already defined here

2023-12-23 16:39:52.571755 |INFO | [rudra-progress] Rudra started
2023-12-23 16:39:52.571921 |INFO | [rudra-progress] SendSyncVariance analysis started
2023-12-23 16:39:52.571935 |INFO | [rudra-progress] SendSyncVariance analysis finished
2023-12-23 16:39:52.571938 |INFO | [rudra-progress] UnsafeDataflow analysis started
2023-12-23 16:39:52.579893 |INFO | [rudra-progress] UnsafeDataflow analysis finished
2023-12-23 16:39:52.579921 |INFO | [rudra-progress] Rudra finished
Error (UnsafeDataflow:/VecSetLen): Potential unsafe dataflow issue in `utils::Digest::to_bytes_len`
-> src/utils.rs:19:5: 24:6
fn to_bytes_len(&mut self, len: usize) -> Vec<u8> {
        let mut bytes: Vec<u8> = Vec::with_capacity(len);
        unsafe { bytes.set_len(len); };
        self.read(&mut bytes[..]).unwrap();
        bytes
    }

Error (UnsafeDataflow:/VecSetLen): Potential unsafe dataflow issue in `utils::PadBlocks::<I, P, F>::fill_buf`
-> src/utils.rs:91:5: 120:6
fn fill_buf(&mut self) -> io::Result<&[u8]> {
        if self.finished {
            if self.buf.len() > self.block_len && self.len > 0 {
                self.len = 0;
                return Ok(&self.buf[self.block_len..self.block_len*2]);
            }
            return Ok(&self.buf[0..0]);
        }
        
        self.buf.clear();
        unsafe { self.buf.set_len(self.block_len); };
        let mut len = try!(self.inner.read(&mut self.buf[..]));
        self.len += len;
            
        if len < self.block_len {
            unsafe { self.buf.set_len(len); };
            let reader: P = (self.padder)(self.len);
            len += try!(reader.read_pad(&mut self.buf));
            //assert_eq!(self.buf.len(), self.block_len);
            self.finished = true;
        }
        
        if len < self.block_len && len > 0 {
            return Err(io::Error::new(
                    io::ErrorKind::BrokenPipe,
                    "unable to read block to end"));
        }
        
        Ok(&self.buf[..self.block_len])
    }

2023-12-23 16:39:52.779513 |INFO | [rudra-progress] cargo rudra finished

$ docker-cargo-rudra ./ ./rudra-report

This is the output of the report

[[reports]]
level = 'Error'
analyzer = 'UnsafeDataflow:/VecSetLen'
description = 'Potential unsafe dataflow issue in `utils::Digest::to_bytes_len`'
location = 'src/utils.rs:19:5: 24:6'
source = """
fn to_bytes_len(&mut self, len: usize) -> Vec<u8> {
        let mut bytes: Vec<u8> = Vec::with_capacity(len);
        unsafe { �[0m�[31mbytes.set_len(len)�[0m; };
        �[0m�[36mself.read(&mut bytes[..])�[0m.unwrap();
        bytes
    }
�[0m"""

[[reports]]
level = 'Error'
analyzer = 'UnsafeDataflow:/VecSetLen'
description = 'Potential unsafe dataflow issue in `utils::PadBlocks::<I, P, F>::fill_buf`'
location = 'src/utils.rs:91:5: 120:6'
source = """
fn fill_buf(&mut self) -> io::Result<&[u8]> {
        if self.finished {
            if self.buf.len() > self.block_len && self.len > 0 {
                self.len = 0;
                return Ok(&self.buf[self.block_len..self.block_len*2]);
            }
            return Ok(&self.buf[0..0]);
        }
        
        self.buf.clear();
        unsafe { �[0m�[31mself.buf.set_len(self.block_len)�[0m; };
        let mut len = try!(�[0m�[36mself.inner.read(&mut self.buf[..])�[0m);
        self.len += len;
            
        if len < self.block_len {
            unsafe { �[0m�[31mself.buf.set_len(len)�[0m; };
            let reader: P = �[0m�[36m(self.padder)(self.len)�[0m;
            len += try!(�[0m�[36mreader.read_pad(&mut self.buf)�[0m);
            //assert_eq!(self.buf.len(), self.block_len);
            self.finished = true;
        }
        
        if len < self.block_len && len > 0 {
            return Err(io::Error::new(
                    io::ErrorKind::BrokenPipe,
                    \"unable to read block to end\"));
        }
        
        Ok(&self.buf[..self.block_len])
    }
�[0m"""

[Qwaz]

Finally a PR for edition 2021 support! Thanks a lot, I'll check this in a few days.

[Qwaz]

Apologies for the long delay. It's in my TODO list, but I've been busy with works and haven't found time to look at it yet.

[nstilt1]

I understand. Since this seems to be a pretty important repo, and since it relies heavily on the current state of the compiler... would it make sense for the compiler team to adopt this repo? It seems like they are pretty up to date with their own changes, but I understand that it would take a while just to update it to a current nightly version. I might give it a go at another update, but it is kind of tedious to find every deleted file and every deleted line to try and figure out which new thing replaced the deprecated things. Do you know if there is a more organized record of the nightly changes aside from the commit history?

[Qwaz]

I'm failing to recreate Rudra-PoC bugs with this PR. I'm investigating if there's any semantic difference between tcx.hir().krate().visit_all_item_likes / tcx.hir().krate().items() and tcx.hir().visit_all_item_likes / tcx.hir().items().

Update: The changes look consistent with rust-lang/rust@db9fea5; I'm not sure why Rudra is affected by this change.

[Qwaz]

To my shame, nightly-2021-08-20 already returns different result from nightly-2020-08-26 (the version used in the paper). I should have tested this version more thoroughly. I'll track this as a separate issue and merge this PR.

[Qwaz]

Do you know if there is a more organized record of the nightly changes aside from the commit history?

I'm afraid to say that I don't know the best way to track this. That's the main reason why it's difficult for me to maintain this repository.

Regarding the suggestion for an adoption, I'm happy to see this forked or maintained by someone else. Rudra project has MIT+APACHE dual license, which should already allow this type of adoption. On the other hand, I'm unsure if it meets the required quality of the compiler team's expectation. There are many rough edges to be polished in Rudra to surpass the "research quality" product 😅